Known diodes such as photosensors or photodetectors have semiconductor layers for producing a detectable change in electric current, by exciting electrons across forbidden bands or band gaps in the electronic structure of the semiconductor material. The electrons are excited by photons received from light or electromagnetic waves from radiation intercepted by the photodetector.
Referring to FIG. 1, one type of photodiode is a High Operating Temperature (HOT) nonequilibrium photodiode such as that disclosed by U.S. Pat. No. 5,016,073 issued to Elliott et al. (the “'073 Patent”). The HOT photodiode 100 has a photosensitive region 102 that absorbs incoming radiation in the form of infrared photons traveling through the photodiode along a pathway shown by arrow 104. The energy from the photons excites the electrons in the photosensitive region 102 across the band gap of the semiconductor material, which reduces resistivity and increases current running through the photodiode, and in turn indicates radiation has been detected.
The photosensitive region or active region or zone 102 is either lightly n-type, or as in this case, lightly p-type doped by ion implantation or in-situ doping in a concentration that is close to the natural or thermally generated majority carrier concentration. With this concentration, extrinsic behavior can be established in the active region 102 when the system is under reverse bias such that the impurities (dopants) contribute more carriers than the number of carriers generated thermally across the energy gap. The n-type dopants establish electronic levels that are closer to the conduction band than the valence band. This leads to indirect generation from the electronic levels that require less energy for an electron to jump from the level to the conduction band than an electron in direct generation from the valence band.
The active region 102 should have the minimum possible number of majority carriers in order to improve the detector's signal-to-noise ratio. This is because radiative and Auger generation-recombination processes are responsible for noise, and are less probable with lower majority carrier concentrations.
The desired majority carrier concentrations are obtained by reducing the number of minority carriers in the active region. A reduction of minority carriers in the space charge balance of the photosensitive region 102 results in a corresponding loss of natural majority carriers leaving a very small minority carrier concentration and a low natural majority carrier concentration that is less then the majority concentration provided by the dopant.
One way to minimize or suppress minority carrier concentrations is to cool the photodetector. This, however, involves complex, bulky and expensive equipment.
One alternative solution, as presented by the '073 Patent, discloses a way to reduce the minority carrier concentration in the photosensitive region 102 on a CdxHg1-xTe (cadmium mercury telluride or CMT) photodiode 100 without the conventional cooling parameters. This is accomplished by placing the photosensitive region 102 between an extraction layer or region 106 and an exclusion layer or region 108. The overall effect of this structure is that minority carriers arc removed at the extraction region or layer 106 and are not resupplied at the exclusion region or contact 108.
The extraction region 106 typically has the opposite majority carrier or conductivity type compared with that of the photosensitive region 102 and forms a p-n junction 116 with the photosensitive region. When a reverse bias is applied, the extraction region 106 extracts minority carriers from the photosensitive region 102 by electrons diffusing to the extraction region from the photosensitive region due to its lower conduction band energy, producing the effect of a “weir” or sink.
On the other side of the photosensitive region 102, the exclusion layer or region 108 is a bulk semiconductor layer that prevents minority carrier injection into the photosensitive region 102. The exclusion region 108 has the same conductivity type as the photosensitive region 102 defining a “pp” or “nn” junction or boundary 114 between the chemically different regions that establish different sizes of energy band gaps. The exclusion region 108 can be degenerately doped to be of the same conductivity type as the photosensitive region; i.e., the exclusion region is n+ or p+ according to whether the photosensitive region is n or p type respectively. In the alternative, the exclusion region is a heterojunction structure provided by a different and wider band gap semiconductor material, than in the photosensitive/active region 102 but with like majority carrier types, forming an nn or pp structure.
Electric fields and large external voltage drops over the exclusion region's length can drive the minority carriers toward the pp or nn junction 114 and into the photosensitive region 102. These are avoided by heavily doping the exclusion region 108. When the electric fields in the exclusion region are reduced enough or eliminated, the minority carriers are transported by diffusion.
The length of the exclusion region 108 is at least three times a minority carrier diffusion length to minimize or prevent the “in-diffusion” of minority carriers from the biasing contact. The '073 Patent discloses that the exclusion layer thickness should be at least 150 μm for doping [NA−ND]>1×1017 cm−3 to prevent diffusion into the active region. More recent, and accurate calculations conclude that three minority carrier diffusion lengths is approximately 60 μm for [NA−ND]=1×1017 cm−3 for p-type, bulk Hg0.7Cd0.3Te at a temperature of 230° K as shown in FIG. 2.
Regardless of this difference, attempting to provide a 60-150 μm thick, bulk semiconductor exclusion region is extremely difficult as using Molecular Beam Epitaxy (MBE) crystal growth processes, as are typically employed for producing photodetectors or photodiodes.
Molecular Beam Epitaxy (MBE) is a chemical vapor deposition method in which a crystal or layered structure is grown on a template (substrate) within a chamber. The substrate is brought to, and kept at, a predefined growth temperature by a heating element typically placed behind the substrate. This is to ensure that sufficient energy is transferred to the substrate's surface to achieve specific reactions.
The structure is grown by providing atomic and/or molecular fluxes obtained by thermal evaporation of the charge materials. The growth process occurs in an ultra-high vacuum environment to minimize the presence of foreign atoms. Polycrystalline and/or amorphous materials are loaded into crucibles and constitute the charges. The fluxes are adjusted by controlling the temperatures of the charge materials. In this way, the incoming atoms/molecules from the charges have to spend a certain residence time on the surface while traveling/diffusing around in order to find a geometrical position that minimizes the surface energy. Shutters between the charges or charge materials in the effusion cells and substrate are individually controllable. They permit/forbid the flow of molecular beams of the particular materials that are desired at a particular time.
Since MBE has a typical growth rate of 10−10 m per second or about 2.5 μm per hour (for growing a 60 μm thick HgCdTe bulk exclusion layer for instance), it requires a growth duration of 24 hours. This long amount of time requires a substantial expense in equipment, maintenance, raw material and labor per exclusion region. This expense per exclusion region could be reduced if the diffusion length of the carriers in the exclusion region could be reduced, which results in a reduced total thickness that is still three times the diffusion length of its carriers.
The MBE extended time period required for growing the relatively thick exclusion layers also causes defects in the growth or growing crystal itself. In the MBE process, solid source materials are evaporated by heating them in effusion cells in order to transport them onto the growing crystal. The process of evaporation changes the surface area, shape and roughness of the source material left behind in the effusion cell. This results in drifts or changes over time in the composition of the fluxes of the material leaving the effusion cell. The variation in material fluxes changes the ratios of the fluxes of the Hg to CdTe to Te sources used to make the crystal (or in other words a variation in the ratios of the flux of material from each effusion cell used). This will vary the composition, and can result in, for example, the establishment of an undesired energy gap. This random variation in energy gap can lead to absorption of infrared radiation in the exclusion layer. The varied composition can also create undesired barriers to carrier flow such that the exclusion layer blocks the flow of majority carriers. Finally, a varied composition can result in poor quality crystals with varying diffusion lengths resulting in unpredictable device performance.
Another problem that can occur in the MBE process of manufacturing thick bulk semiconductor exclusion layers is that while the crystal is thickened, it absorbs more infrared radiation from the substrate heater resulting in a rise in temperature over time. The power to the substrate heater is ramped down in order to compensate for the rise in temperature. However, the ramping down of the power frequently cannot be perfectly matched to the temperature rise because conflicting temperature requirements exist that are associated with emissivity as the crystal grows and require temperature durations not related to the rise in temperature. These variations in growth temperature lead to variations in material composition, and in turn, variations in crystal quality.
In another alternative, disclosed by U.S. Pat. No. 6,081,019 issued to White and fully incorporated herein, and which is illustrated in FIG. 3, a HOT photodiode 120 is provided with a buffer layer 128 between the exclusion layer 124 and the active layer 122, and a buffer layer 130 between the extraction layer 126 and active layer 122. The layers are disposed on a growth substrate 136 and are biased through contacts or electrodes 132, 134.
Each buffer layer 130, 128 acts as an additional exclusion and extraction layer according to the side (exclusion side or extraction side) of the active layer they are disposed on. The buffer layers 130, 128 have lower dopant concentrations than their corresponding extraction or exclusion layer, although the CMT chemical composition of the layers is otherwise the same as the corresponding extraction or exclusion layer. The chemical composition of the active layer is different than that in the buffer layers so that the buffer layers provide a wider band gap than that in the active layer and with much lower (10 to 100 times) concentration of minority carriers. This results in two extraction interfaces 140 on either side of the extraction buffer layer 130 and two exclusion interfaces 138 on either side of the exclusion buffer layer 128. The buffer layers 128, 130 provide a low minority carrier concentration at low doping regions close to the junctions, preventing high thermal generation leading to leakage current.
The “buffered” diode in the '019 patent, however, has a total composite exclusion semiconductor region (126 and 130) thickness of 7.0 μm, which is substantially below the thickness of three times the diffusion length of minority carriers (approximately 60 μm for CMT) which is desired for high quality bulk material crystals. Such thin layers accordingly sacrifice quality and performance by permitting unacceptable levels of minority carriers to diffuse into the active layer.
In addition, while the buffer layers 128, 130 are actually “buffering” the dopants themselves (i.e. the dopant atoms) and provide for a relatively thicker exclusion region to counter some diffusion length, the composition of the buffer layers 128, 130 do not provide a stronger physical barrier to minority carriers. In other words, the same general proportion of minority carriers that would diffuse from the exclusion layer 124 and into the active region 122 would also diffuse from the buffer layer 128 if they occupied the same position relative to the biasing contact.